Instruction-set matching and GA-based selection for embedded-processor code generation

Instruction Selection

“…To solve the pattern selection problem, Shu et al [308] employed the theories of genetic algorithms (GAs), which mimic the process of natural selection (see for example [292] for an overview). 11 The idea is to formulate a solution as a string, called a chromosome or gene, and then mutate it in order to hopefully end up with a better solution.…”

“…Seemingly independently from the earlier work by Shu et al [308] (discussed in Chapter 3), Lorenz et al [241,242] The design by Lorenz et al is basically an iterative process. First, the operations within a block are scheduled using list scheduling, which is a traditional method of scheduling (see for example [290]).…”

Instruction-set selection for multi-application based ASIP design: An instruction-level study

“…From a cost and performance perspective, types of instructions used in given applications is vital. Approaches to instruction set generation for an ASIP can be classified as either instruction set synthesis approach [7,11] or instruction set selection approach [3,12,16,19] on the basis of how the instructions are generated. In synthesis approach, instruction set is synthesised for a particular application based on the application requirements, while in selection approach, a superset of instructions is available and a subset of them is selected to satisfy the performance requirements within the architectural constraints [13].…”